Abstract: To address the excessive reliance on on-site experience, low matching accuracy between fault characteristics and handling measures, and delayed decision-making responses in stuck pipe incident handling, this paper proposes an intelligent generation method for handling measures based on semantic retrieval and graph reasoning. First, multi-source drilling engineering data are integrated to construct a stuck pipe knowledge graph encompassing fault characteristics, formation parameters, operational parameters, handling measures, and application effects, thereby enabling structured organization and relational representation of domain knowledge. On this basis, a five-step retrieval process of "segmentation–matching–recognition–retrieval–generation" is proposed: the user's unstructured fault description is segmented into semantic sentences; BGE-M3 vector encoding and Chroma indexing are used for sentence-node semantic matching; the stuck pipe type is inferred via hybrid classification combining feature engineering and a large language model; multi-hop retrieval in the Neo4j graph yields a structured handling subgraph, which is then fed into a large language model to generate handling recommendations. In addition, a three-dimensional evaluation model encompassing "operational rationality, hierarchical rationality, and completeness" is established for quantitative assessment and prioritized recommendation of the generated measures. Experimental results on 55 field stuck pipe cases demonstrate that the compliance rate of the generated handling measures reaches 94.5%, representing a 29.0 percentage point improvement over the pure large language model approach, thereby verifying the method's effectiveness and engineering practicality. The deep integration of semantic retrieval and graph reasoning effectively reduces the reliance on on-site experience in stuck pipe handling, improves matching accuracy, and enhances decision-making response efficiency, providing a knowledge-traceable and field-applicable technical solution for intelligent emergency management in drilling engineering.